A combined Bayesian framework for satellite reliability estimation

Reliability estimation for high reliability, long-life complex satellite system has been recognized as a critical and difficult mission for satellite reliability engineer. The information for satellite reliability estimation is insufficient at the system level. The available information is at various levels of the system in different types such as the pass/fail data, lifetime data, degradation data, subjective knowledge and experience by experts and so on. To thoroughly integrate the heterogeneous multilevel sources of information, this paper adopts the Bayesian methodology and other methods for complex system reliability estimation from the perspective of multilevel information fusion. By utilizing these methods, a systematic framework for satellite reliability estimation is constructed. The estimation is carried out from the component level, subsystem, and then rolling up, to system level. The components and subsystems are categorized into three groups according to the available information and the specific reliability estimation model for each group respectively. The satellite system is modeled based on the reliability diagram consisting of all the subsystems and components. Hierarchy Bayesian modeling is adopted to combine all kinds information for system reliability estimation. We illustrate the methodology through an example and demonstrate that the satellite reliability estimation framework can handle the reliability estimation with heterogeneous multilevel information properly.